A Comprehensive Radiomics Model for Predicting Progression-Free Survival in High-Risk Neuroblastoma Based on CT Imaging.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
PURPOSE: This study evaluates the predictive utility of clinical features, CT imaging characteristics, radiomic features, and multimodal statistical models in estimating progression-free survival (PFS) in high-risk neuroblastoma (NB) patients. MATERIALS AND METHODS: 105 pathologically confirmed high - risk NB patients per COG criteria were retrospectively selected and divided into training (80%) and test sets (20%). Statistical models were built using clinical variables, CT features, and radiomic features, with SVM algorithms employed. Five models were assessed, including radiomics, clinical + CT, clinical + radiomics, imaging + radiomics, and combined clinical + CT + radiomics models. A nomogram was constructed to predict PFS. RESULTS: Univariate analysis found NSE > 200 μg/mL, LDH >1000 U/L, and MYCN amplification as significant poor - prognosis predictors (P < 0.05). Multivariate analysis confirmed MYCN amplification (P = 0.043) and IDRFs presence (P = 0.036) as independent risk factors. A radiomics model was built from CT scans' different phases. The combined model showed the best and most stable performance with AUC values of 0.855, 0.796, 0.856 in training, validation, and test cohorts. A nomogram integrating Rad - Score and clinical parameters from Cox regression was developed, with a C - index of 0.806 (P = 0.028) via bootstrap validation. CONCLUSION: This study validates a PFS predictive model combining clinical, CT, and radiomic features, providing a robust tool for individualized prognosis in high-risk NB patients. ADVANCES IN KNOWLEDGE: Radiomics demonstrates predictive utility for progression-free survival in pediatric patients with high-risk neuroblastoma.